Cellular heterogeneity plays a critical role in many disease states. The cancer cell model suggests that a small subset of tumor initiating cells (TIC) or cancer stem-like cells (CSC) contribute cancer heterogeneity in many tumors, and are necessary to initiate and sustain cancer growth. They are resistant to traditional therapy and capable of division and differentiation to give rise to a heterogeneous population of tumor cells. In both bulk and CSC populations, there is a high degree of heterogeneity: Phenotypic heterogeneity arises in the bulk populations from differentiation and regulation, while genetic mutations in the CSCs create multiple stem-like subpopulations. Despite its importance, only small advances have been made in characterizing this multi-level heterogeneity or how it is regulated. Tools such as petri dishes are incapable of studying these phenomena as they can only assess the average response of large numbers of cells. FACS techniques allow analysis of population distributions, but provide little insight into intra-population heterogeneity. There is clear need to provide tools for single cell handling and characterization of these rare but critica subtypes to dissect the various factors contributing to cancer heterogeneity. This grant will focus on the development of microfluidic platforms to examine CSC heterogeneity by integrating multiple innovative approaches for single cell derived sphere formation when simultaneously co-cultured with adherent stromal cells, followed by cell and sphere retrieval for single-cell resolution multiplexed gene expression analysis using the Fluidigm C1/Biomark HD platform. The proposed platform and assays provides marker-free CSC identification and investigation of microenvironmental control of stemness, allowing the concurrent characterization of mutation-based (genetic) and exogenously-regulated (phenotypic) heterogeneity in CSCs. Specific Aims: In Specific Aim 1, we will develop and optimize microfluidic co-culture platforms for growth, retrieval, and expression analysis of breast cancer spheres. We will develop our co-culture microfluidics for high-throughput (> 1,000 wells), optimize a robust methodology for retrieval and dissociation of the spheres, and validate the loading of spheres into the Fluidigm C1/Biomark HD for multiplexed gene expression analysis. In Specific Aim 2, we will use the developed co-culture platform to characterize heterogeneity in single cell- derived spheres in breast cancer cell lines when co-cultured with normal and cancer-associated stromal cells in engineered microenvironments. Finally, in Specific Aim 3, we will analyze CSC heterogeneity and differentiation in primary tissue samples. Results observed in the breast cancer cell lines (Aim 2) will be compared and validated in patient derived xenografts (PDXs) of three subtypes: basal, HER2+, and luminal breast cancer patient samples.